Stephen Brooks 2003-09-25 09:04:52 | My workstation arrived on Wednesday 17th, so basically I've spend 6 proper working days on the Muon code so far. There were an unbelievable number of bugs when I recompiled it under WinXP! I've fixed several buffer-overruns and some memory-related glitches, plus I think an intermittent bug in the checksum handling caused by floating-point rounding. So we are currently heading for v4.32, along with a new optimisation, called SolenoidsTo15cm, which will actually have some more serious science value than the current job that is running. I'll use the results from that to complete a report I'm meant to be writing on this (which probably will be published, at least as an RAL technical report, and quite possibly a paper in the Journal of Physics G: Nuclear and Particle Physics). Once that job is running, I might be able to turn my attention in the direction of better and/or more reliable stats. In the mean time, I attach a spreadsheet of some benchmarking I did on this workstation, to see if hyperthreading benefits Muon1. Result: it does! A 40-45% increase in speed was got here over SMP alone. You can also see stacks.iop.org/JPhysG/39/1673 for a paper; not written by me, but I'm on the authors list as it references to my work. And Figure 3 is taken from a variant of Muon1 I used for tracking individual particles' paths. It doesn't make any sense: that's why they call it "virtual" |
Stephen Brooks 2003-09-26 13:59:10 | Here we go... v4.32 is up and I hope operational. Stats may be a bit flakey for a while yet but their time will come. Details of the upgrade are here. It doesn't make any sense: that's why they call it "virtual" |
[DPC]TeamNWW - Huub 2003-09-27 04:30:08 | [DPC]TeamNWW is now running 4.32 on several machines (4 to be exact).. more will be added after the weekend ps. adding muon to my startup folder always produces a command-box. i tried shortcut to the (edited) background batchfile and a fresh shortcut with the -b option.. |
Stephen Brooks 2003-09-27 05:30:52 | Firstly just a quick post to say "you guys are excellent!" I only posted the new version late last night and when I wake up this morning there are already >350KB of results in the database from about 7 users. This afternoon I'll try and get a regular best100-generation up and running, so we have some genuine distributed computing going on.quote:OK, yeah, that was the one 'known issue' with 4.32 that I found out first: the -b switch worked on WinXP but not 98. However, I think it's easy to fix. Can you tell me the text in the title of those commandboxes? (So I can be sure to fix it for you as well?) Judging from what's happening on another thread there's a divide-by-zero and one other immediate-crash in there too. It doesn't make any sense: that's why they call it "virtual" [This message was edited by Stephen Brooks on 2003-Sep-27 at 13:43.] |
[DPC]TeamNWW - Huub 2003-09-27 09:52:09 | It just says "Muon".. |
Stephen Brooks 2003-09-27 10:01:24 | Mine here says "MUON1" so I'll use that. It doesn't make any sense: that's why they call it "virtual" |